Massimiliano
Cavallini
*a,
Felice C.
Simeone
a,
Francesco
Borgatti
a,
Cristiano
Albonetti
a,
Vittorio
Morandi
b,
Claudio
Sangregorio
cd,
Claudia
Innocenti
d,
Francesco
Pineider
d,
Emilia
Annese
e,
Giancarlo
Panaccione
e and
Luca
Pasquali
f
aCNR-ISMN, Via P. Gobetti 101, I-40129, Bologna, Italy. E-mail: m.cavallini@bo.ismn.cnr.it
bCNR-IMM Bologna, Via P. Gobetti 101, I-4012, Bologna, Italy
cCNR-ISTM Milano, Via C. Golgi 19, I-20133, Milano, Italy
dINSTM and Dep. of Chemistry “ U. Schiff” Università di Firenze, Via della Lastruccia 3–13, 50019, Sesto Fiorentino (FI), Italy
eCNR-INFM-TASC S.S. 14-Km 163.5, AREA Science Park, I-34149, Basovizza (TS), Italy
fDep. of Materials and Environmental Engineering, University of Modena and Reggio Emilia, Via Vignolese 905, 41100, Modena, Italy
First published on 9th August 2010
We present a novel additive process, which allows the spatially controlled integration of nanoparticles (NPs) inside silicon surfaces. The NPs are placed between a conductive stamp and a silicon surface; by applying a bias voltage a SiO2 layer grows underneath the stamp protrusions, thus embedding the particles. We report the successful nanoembedding of CoFe2O4 nanoparticles patterned in lines, grids and logic structures.
Among the several methods and processes proposed for the integration of nano-objects in devices,10–13 the ones relying on the “embedding” of materials of interest are largely used for applications in optics,14,15 micro-mechanics16 and magnetism17,18 to better exploit the functionalities of nano-objects or to improve some physical properties of the matrix. In conventional embedding nano-objects are usually dispersed inside a protective environment established by a matrix which can either be organic or inorganic to exploit the properties of the individual objects, otherwise hampered by possible chemical instabilities or aggregation phenomena. Up to now, several different embedding procedures have been developed,10–13 but some unresolved issues still need to be addressed. In particular, no spatial control of dispersed nano-objects can be achieved unless embedding is combined with complex and expensive lithographic procedures.19
In this work we present a novel approach named nanoembedding (n-EM) which merges in a single process the advantages of a spatially controlled nanopatterning20 and the conventional embedding approach. We report here the successful nanoembedding of cobalt ferrite (CoFe2O4) nanoparticles with average diameter 6.0 ± 1.3 nm (see ESI†) on a silicon wafer. CoFe2O4 is a largely investigated material which, thanks to the large magnetic anisotropy,21 exhibits high blocking temperatures even for very low grain size. This property, jointly with its chemical stability, makes CoFe2O4 an appealing magnetic material for several applications such as high-density information storage, high performance small components (e.g. capacitors), anti-reflection coatings and catalysis.22 A scheme of the process is depicted in Fig. 1: the nano-objects are placed between the protrusions of a conductive stamp and a semiconductive surface in a vapor rich environment. By applying a bias voltage, the sample surface is oxidized, forming an oxide film that grows around the nano-objects. The process acts only under the stamp protrusions, therefore it does not require any lithographic manipulation of the nano-objects, which can be directly deposited either on the stamp or onto the surface (see ESI†).
Fig. 1 Scheme of nanoembedding. (a) NPs are deposited on the surface. (b) A conductive tip, or stamp protrusion, is placed in proximity of the NP in a vapour rich environment that forms a meniscus, after that a bias is applied, oxidizing the surface. (c) The NPs under the stamp protrusion are embedded into the oxide coating. (d) Non-embedded NPs are removed by washing the sample. |
The thickness of the capping oxide layer can be varied between 2 and 25 nm and depends on the bias voltage, application time, size of the nano-objects and the relative humidity (RH), similarly to the conventional local oxidation technique.23–25 Using 6 nm size CoFe2O4 NPs, 20 V bias applied for 10 s with RH 80% we obtained a SiOx thickness of 10 nm. After the n-EM procedure, the non-embedded objects can be easily removed by washing the sample surface with the same solvent used for NPs deposition (this step is not necessary if the NPs are deposited on the stamp).
The application of n-EM to real systems is shown in Fig. 2. Here the stamp motif consists of a series of parallel lines 200 nm wide with a 1.5 µm periodicity (Fig. 2a–d) and a logic pattern, i.e. a pattern containing information in binary code (Fig. 2e, see also ESI†).
Fig. 2 (a) Scheme of 2-step nanoembedding; (b) 60 × 40 µm2 SEM micrograph of CoFe2O4 NPs embedded on silicon oxide inside parallel stripes; (c) corresponding AFM (zoom) image; bar is 1 µm, Z scale 0–20 nm; (d) results of 2-step n-EM obtained from sample b after repeating the process rotating the same stamp by 90°; bar is 1 µm, Z scale 0–20 nm; (e) single stripe of CoFe2O4 NPs embedded at high density of NPs (∼300 NPs µm−2); bar is 200 nm, Z scale 0–20 nm; (f) demonstration of a logic pattern obtained from the replica of a digital video disk; bar is 3 µm, Z scale 0–25 nm. |
Fig. 2b and c show SEM and AFM images of parallel stripes containing a low density of NPs (∼75 NPs µm−2, measured from AFM image) while Fig. 2d shows the same sample in which n-EM has been performed twice, rotating the stamp 90° before the second application (2-step process). The images clearly show the repeatability of n-EM on the same sample and the possibility to obtain complex distributions by simply combining multiple applications additively.
The quantity of embedded NPs is controlled during their deposition on the stamp (or on the surface) controlling the concentration of the initial solution. As an example, Fig. 2e shows the detail of a single stripe obtained using a higher density of NPs (∼300 NPs µm2 measured from AFM images). Finally, Fig. 2f shows the patterning of a logic structure containing dots and lines which are the basic elements to fabricate a great number of complex structures. The process was successfully applied over a 1 × 1 cm2 area.
The chemical and magnetic features of the nanoembedded NPs were characterized by X-ray Absorption Spectroscopy (XAS), X-ray Photoelectron Spectroscopy (XPS) and by Superconducting Quantum Interference Device (SQUID) magnetometry to ensure that the specific functionalities of the CoFe2O4 NPs are not influenced by the n-EM.
Fig. 3a and b show the X-ray core-level absorption spectra of the Fe and Co L2,3 thresholds for CoFe2O4 NPs embedded on silicon (both 1- and 2-step processes), compared to a film deposited by drop casting. The spectra were normalised to maximum peak height after the subtraction of a constant offset for the background. The L2,3 adsorption edges exhibit strong resonances related to the 2p63dn (ground state) to 2p53dn+1 (excited state) dipole-allowed transitions, which provide a detailed picture of the local electronic structure of these elements.26 The structured lineshape of Fe and Co L2,3 spectra intimately reflects the chemical state and the local coordination symmetry of these atomic species. All XAS spectra exhibit the typical multiplet splitting fine structure of CoFe2O4.27 The comparison between our spectra and the literature results20 provides immediate indication that the CoFe2O4 cores of the NPs are unaffected by the chemical processes occurring during the embedding, however, we do not exclude a possible partial chemical reduction (or oxidation) of NPs that are in contact with the stamp during the bias application, in fact, occasionally (in not optimized samples, with very low coverage of NPs), we observed the presence of Fe2+ at XPS spectra due to a partial electrochemical reduction of NPs.
Fig. 3 Characterization of embedded nanoparticles. (a) XAS spectra of Fe and (b) Co L2,3 absorption thresholds for embedded NPs compared with spin coated NPs thick film. (c) Si2p photoelectron spectra collected with photon energy 1140 eV over different positions of a sample surface with NPs embedded by 2-step process. |
XPS analysis has been employed to probe the chemical state of the other elements of the surface. The species detected at the surface by XPS were Si (Si2p), O (O1s), and C (C1s). The O1s and C1s signals are not very meaningful because of the overlap of several contributions from the embedded material, residuals of the solvent used for the deposition process and surface contaminants due to air exposure before insertion in the UHV chamber (e.g. CO2, H2O, hydrocarbons, etc.). Conversely, the Si2p signal is only related to the silicon substrate and distinguishes clearly between valence states associated to different chemical species. The Si2p XPS spectrum shown in Fig. 3c was recorded for a sample surface with the NPs embedded by a 2-step process. The spectrum exhibits two strong peaks which are related to oxidized silicon distributed over the surface (Si4+, B.E. 104 eV, indicating the presence of SiO2) and to the underlying bare silicon (Si0, B.E. 100 eV), respectively. XAS and XPS spatial mappings of the surfaces were performed for all the samples. The branching ratio of the two signal intensities is preserved over the whole surface, thus indicating that after the embedding process the silicon oxide distribution is homogeneous. Slight variations among the Si2p spectra are reasonably explained by the minor presence of silicon chemical states related to different oxide stoichiometry28 and to surface contamination due to exposition to the air, being the samples prepared ex situ and analysed without any surface treatment after the transfer in vacuum for XPS and XAS measurements. Therefore these results support the ability of the embedding process to act effectively over macroscopic large-areas.
The magnetic characterization of embedded NPs was performed using a SQUID magnetometer. In Fig. 4 is reported the temperature dependence from 5 to 300 K of the zero field cooled (ZFC) and field cooled (FC) magnetization curves of as prepared NPs, spin coated and embedded NPs respectively. The curves were obtained with a 10 mT magnetic field that in the case of spin coated and embedded NPs was applied perpendicularly to the silicon slab. All samples display the typical behavior of single domain magnetic NPs: at room temperature all particles are in the superparamagnetic regime, consistent with the very small size of the NPs, while below ca. 300 K, thermal irreversibility occurs. No alteration of the magnetic properties was observed in embedded NPs with respect to the spin coated ones, whose behavior indeed resembles those of powder samples of non-deposited NPs (Fig. 4a). Eventually, it is worth to point out that no differences in the magnetic properties were observed when applying the magnetic field in the plane or perpendicularly to the plane of the slab. These results support the ability of the embedding process to act effectively over macroscopic large-area regions without alterations of magnetic properties in the embedded NPs.
Fig. 4 Temperature dependence of the zero field cooled (ZFC, open symbols) and field cooled (FC, full symbols) magnetizations of embedded NPs (A, red curves) compared to spin coated thin deposits (B, blue curves). The inset shows the magnetizations of as prepared (powder) NPs. |
Footnote |
† Electronic supplementary information (ESI) available: Detailed method and characterization. See DOI: 10.1039/c0nr00315h |
This journal is © The Royal Society of Chemistry 2010 |